Know, don't believe.
Nyalai returns binary verdicts on probabilistic systems in quantitative finance. Every candidate is submitted to five adversarial validation gates. Every verdict is signed, timestamped, and publicly citable. Institutional model risk management addresses the first three quadrants of decision-making. Nyalai closes the fourth. The methodology is published in full under Creative Commons Zero. The signature is not.
Under Rule 6.8, this verdict is publicly citable and versioned.
View full audit trailOne system passes. One system does not. Both are examined against the same five gates, under the same Doctrine v0.3.3.1, and both receive a signed audit trail. Nyalai does not grade on a curve. The Doctrine either binds, or it does not.
Illustrative demonstrations. Verdict-form 001 metrics on Synapse v0.3 are MEASURED against a 239-prediction ledger. The CALIBRATED example is constructed for pedagogical clarity and does not represent any specific live verdict. Full audit trail for the NOT-CALIBRATED case at nyalai.com/verdicts/001-synapse-v03.
Quantitative finance produces published strategies at industrial scale. Very few of them survive an honest out-of-sample test. The statistical inference horizon in financial time series is short relative to the horizon over which the underlying claims need to hold, and Fama and French have documented that even ten-year windows admit substantial probability of misleading realized premia. Nyalai exists because the gap between what is published and what is real has become large enough to be its own asset class, and because internal review cannot solve the horizon problem alone.
In the taxonomy Nassim Taleb published in 2008, decisions divide along two axes: whether the payoff structure is simple or complex, and whether the underlying randomness lives in thin-tailed or fat-tailed distributions. Quantitative strategies operate in the complex-and-fat-tailed region: nonlinear payoffs, unbounded loss possibilities, regime shifts. SR 11-7 and equivalent model risk management guidance address the first three regions adequately. The fourth is where standard validation methodology fails, and where an external adversarial layer is not an addition. It is a requirement.
Nyalai does not eliminate Fourth Quadrant risk. Nyalai makes the decisions rendered under that risk auditable, and refuses strategies whose evaluation would require thin-tailed assumptions the underlying data does not support. The five gates are the operational form of that discipline. Each gate uses the statistical framework appropriate to its question, following the modular validation principle Joseph Simonian formalized in the Journal of Financial Data Science.
A refused system (our own meta-calibration engine, Synapse v0.3) side by side with an approved system (a reference case). The rust wax seal signals refusal. The green wax seal signals approval. A reader understands the mechanism in five seconds.
A shared vocabulary for how much trust to place in a probabilistic prediction system. Numbered ascending. Each level defined by empirical passage of specific gates. Each level attached to a publicly declared deployment context. Strict nesting. Binary per level. Publicly contestable. Time-bounded.
Every verdict is a long-form artifact. Signed. Versioned. Archived. Publication cadence: bi-weekly starting September 2026. Doctrine documents, method notes, and case studies live alongside.
A meta-calibration engine developed by the same author is submitted to CalibrationJudge on a ledger of 239 resolved predictions. Every gate fails at published thresholds. The first Nyalai Verdict-form is a refusal of our own system. That is the intended outcome of an author who takes the sovereignty rule seriously.
